Mammalian signaling networks comprise biochemical pathways with shared components, common inputs, and overlapping outputs. Understanding how information flows through these pathways requires information on signaling networks as a whole, rather than on one or two components. To study signaling at a systems level, we need ways to measure the abundance and post-translational modification of many proteins in a parallel, quantitative, and reliable manner. In addition, since an understanding of signaling requires the frequent temporal sampling of many proteins under multiple conditions, these methods must be high-throughput. Here, we describe technology that mimics an immunoblot, but in a multiplexed and extremely miniaturized format. Cells are cultured in 96-well plates and subjected to a variety of perturbations (stimulation with epidermal growth factor in the presence of selected shRNA's or cDNA's). The cells are then lysed and the lysates arrayed at high spatial density onto glass-supported nitrocellulose pads, also arranged in a 96-well format. By probing each pad with a different antibody, the `state'of the signaling network is assessed. Currently, high- throughput multidimensional readouts can be obtained either by automated fluorescence microscopy or by multiplexed flow cytometry. Although both techniques provide the ability to track more than one protein simultaneously, they rely on the use of different colored fluorophores and hence can only follow about a dozen proteins. In contrast, the technology described here enables a single sample to be replicated thousands of times on separate microarrays and is thus easily scaled. This application details efforts to make lysate microarray technology rigorously quantitative and outlines automation strategies that render it high-throughput and reproducible. In addition, since one of the biggest challenges in analyzing cancer at a systems level is to go beyond a mere description of the data, a strategy is also presented to build predictive models of cell signaling using Bayesian methods. As proof-of-concept, we will focus on epidermal growth factor signaling in A431 cells. Although this system is relatively well-understood, our approach should capture higher-order interdependencies between proteins that are not evident from traditional studies. More importantly, our strategy should provide a general way to uncover causal relationships in less well-studied networks using data derived from our high-throughput microarrays.